Supplementary data for "BCDDM: Branch-Corrected Denoising Diffusion Model for Black Hole Image Generation"

We release the data products associated to the paper BCDDM: Branch-Corrected Denoising Diffusion Model for Black Hole Image Generation. The data can be used in conjunction with the code BCDDM to reproduce the results of the paper. This dataset contains 2,157 simulated black hole images generated via General Relativistic Radiative Transfer(GRRT) simulations of Radiatively Inefficient Accretion Flows (RIAF), using the `ipole` code. The dataset serves as training data for the Branch-Corrected Denoising Diffusion Model (BCDDM) [arXiv:2502.08528], enabling physics-conditioned black hole image generation through seven key physical parameters of the RIAF model.本团队发布与论文《BCDDM：用于黑洞图像生成的分支校正去噪扩散模型》相关的数据产品。该数据集可配合BCDDM代码复现该论文的实验结果。本数据集包含2157张模拟黑洞图像，这些图像通过针对辐射低效吸积流（Radiatively Inefficient Accretion Flows，RIAF）的广义相对论辐射转移（General Relativistic Radiative Transfer，GRRT）模拟生成，生成过程采用`ipole`代码。该数据集作为分支校正去噪扩散模型（Branch-Corrected Denoising Diffusion Model，BCDDM）[arXiv:2502.08528]的训练数据，可通过辐射低效吸积流模型的七个关键物理参数，实现基于物理约束的黑洞图像生成。